treecutter/sitemap.py

   1 #!/usr/bin/python
   2 import os
   3 import re
   4 import time
   5 import shutil
   6 import gettext
   7 import tempfile
   8 from amara import bindery
   9 from treecutter.trie import Trie
  10 from treecutter.link import Link
  11 from treecutter.tools import ssh_cmd, publish, mkdir_p
  12
  13 class Sitemap():
  14     """Class keeping the internal site structure"""
  15     def __init__(self):
  16         self._file = 'sitemap.txt'
  17         self._tree = Trie()
  18         self._sitelang = set()
  19         self._isocode = bindery.parse('/usr/share/xml/iso-codes/iso_639_3.xml')
  20         self._tranlang = {}
  21
  22     def add_link(self, link):
  23         tokens = filter(None,re.split(r'(^/[\w-]*/|[\w-]*/)',link))
  24         self._tree.add(tokens,Link(link))
  25
  26     def write_map(self):
  27         f = open(self._file,'w')
  28         f.write('\n'.join(link.link() for link in self._tree))
  29         f.close()
  30
  31     def read_map(self):
  32         try:
  33             f = open(self._file)
  34             sml = f.read().split()
  35             f.close()
  36             for line in sml:
  37                 self.add_link(line)
  38         except IOError, what_error:
  39             print 'INFO: Could not read sitemap.txt - one will be created'
  40
  41     def set(self):
  42         return set(link.link() for link in self._tree)
  43
  44     def process(self):
  45         t1 = time.time()
  46         for link in self._tree:
  47             link.prepare()
  48         t2 = time.time()
  49         print "Prepare  [%5.2f s]" % (round(t2-t1,2))
  50         for link in self._tree:
  51             self._sitelang = self._sitelang.union(set(link.languages()))
  52         for tran in self._sitelang:
  53             if tran != 'en':
  54                 self._tranlang[tran] = gettext.translation('iso_639_3',
  55                                                            languages=[tran])
  56         t3 = time.time()
  57         print "Language [%5.2f s]" % (round(t3-t2,2))
  58         for link in self._tree:
  59             link.render()
  60         t4 = time.time()
  61         print "Render   [%5.2f s]" % (round(t4-t3,2))
  62         for link in self._tree:
  63             link.template(self)
  64         t5 = time.time()
  65         print "Template [%5.2f s]" % (round(t5-t4,2))
  66         t6 = time.time()
  67         res = set()
  68         cwd = os.getcwd()
  69         for link in self._tree:
  70             res = res.union(link.resources())
  71         for f in res:
  72             outfile = tmptarget+f
  73             mkdir_p(os.path.dirname(outfile))
  74             shutil.copyfile(f,outfile)
  75         print "Resources[%5.2f s]" % (round(t6-t5,2))
  76         sitmaplink = Link('/sitemap')
  77         for l in self._sitelang:
  78             sitmaplink.add_page((l,'/sitemap.'+l+'.xml'))
  79         for l in self._sitelang:
  80             sitmaplink.page(l).set_article(self.gen_menu(l,None,"tree sitemap"))
  81             sitmaplink.page(l).template(self)
  82         t7 = time.time()
  83         print "Sitemap  [%5.2f s]" % (round(t7-t6,2))
  84
  85     def graph(self):
  86         self._tree.graph()
  87
  88     def gen_menu(self,lang,page,cssclass):
  89         return self._tree.menu(lang,page,cssclass)
  90
  91     def lang_menu(self,lang,link):
  92         html = "<ul>"
  93         for l in link.languages():
  94             isoxml = u"//iso_639_3_entry[@*='"+l+"']"
  95             ln = self._isocode.xml_select(isoxml)[0].name
  96             if lang != 'en':
  97                 ln = self._tranlang[lang].gettext(ln)
  98             p = link.link()
  99             if p[-1] == '/':
 100                 p = p +'index'
 101             p = p+'.'+l
 102             html += '<li><a href="%s" hreflang="%s">%s</a></li>' % (p, l, ln)
 103         html += "</ul>"
 104         return html
 105
 106     def publish(self):
 107         ssh_cmd(args.output,"mkdir -p")
 108         publish(tmptarget, args.output)
 109         for res in ["css","images","js","favicon.ico"]:
 110             if (os.path.exists(args.style+res)):
 111                 publish(args.style+res, args.output)
 112         ssh_cmd(args.output,"chmod a+rx")